Over the last decade or so, we have witnessed a surge in the number of ‘Writing and Editing Software’ available in the market. These “tools” have gone from being relatively simple ‘spelling & grammar checking programs’, to becoming complex ‘AI-Enabled Writing Assistants’ that use machine learning, deep learning and natural language processing to analyse and predict what we want to say and how we should say it. They come in both free and premium versions; and they can be found, in one form or another, everywhere from our mobile phones, to our tablets and computers. They are heavily advertised, often with endorsements by celebrity authors who even go so far as to share how well these applications ‘evaluated their writing’ and ‘facilitated their editing and revision’, and how they are ‘incredibly helpful tools for writers at every level’.
But are these applications truly helpful for writers at ‘every’ level?
In 2013, the Pew Research Centre published the results of its survey (conducted between March 7 and April 23, 2012) of 2,462 Advanced Placement (AP) and National Writing Project (NWP) teachers in the U.S., Puerto Rico and the U.S. Virgin Islands, in its article The Impact of Digital Tools on Student Writing and How Writing is Taught in Schools.
Although some of the responses were positive, some were rather alarming:
“68% said that digital tools make students more likely—as opposed to less likely or having no impact—to take shortcuts and not put effort into their writing.”
“46% (majority) said these tools make students more likely to ‘write too fast and be careless’.”
“40% (majority) said digital technologies make students more likely to ‘use poor spelling and grammar’.”
In addition to these were some worrisome observations of the teachers, such as:
“the increasing tendency of some students to use informal language and style in formal writing assignments”,
“the increasing need to educate students about writing for different audiences using different “voices” and “registers””, and
“the general cultural emphasis on truncated forms of expression….hindering students' willingness and ability to write longer texts and to think critically about complicated topics”.
A year later, on April 15, 2014, Kathleen A. Bronowicki submitted a related study titled Technology’s Adverse Effects on Student’s Writing: An Emphasis on Formal Writing is needed in an Academic Curriculum to the Department of Education and Human Development, at The College of Brockport, SUNY (State University of New York).
According to the study, Robert Rothman, then senior fellow at the Alliance for Excellent Education and now Senior Editor at the National Center on Education and the Economy, Washington, was of the view that “Professors of colleges and authority figures within the career world are seeing a sharp decline in individuals’ writing skills.” (Rothman, 2012, as cited in Bronowicki, 2014, p. 8).
Kathleen argued that “students take advantage of the spell-check and grammar-check options to ensure that their work is free of spelling, grammatical, and/or sentence structure errors. Although these options save students’ time and energy, they lower students’ mental energy and decrease their persistence; furthermore, students have become so accustomed to taking shortcuts and writing in small bites that they fail to see the purpose of writing, editing, revising, and re-writing, let alone using grammatical precision and appropriate sentence structure.” (Bronowicki, 2014, p. 6).
Kathleen’s views resonated with those of Kyle Dodson, an English Language Arts teacher at Clovis West, who, in an interview with The Feather Online, said “It (the software) is there to help you, but once you start relying upon it to solve your problems, it creates problems.” (Parker, 2018).
Expanding our visual field by focusing beyond student life to life in the corporate world, we see a domino effect of this problem on employability and economic efficiency. According to the National Association of Colleges and Employers (USA), “73.4% of employers want a candidate with strong written communication skills.” (Moore, 2016). And, because most of the candidates that apply for jobs these days do not possess them, businesses end up spending billions a year in “remedial writing training” for employees (Moore, 2016).
In his dissertation titled Write to Work: The Use and Importance of Writing as Perceived by Business Leaders, Clay Aschliman, Virginia Commonwealth University, documented and reviewed the data of the study “Writing: A Ticket to Work...Or a Ticket Out”, conducted and published by the National Commission on Writing (NCW) and College Board in 2004. The study surveyed human resources directors of 120 Business Roundtable organizations - a group of chief executive officers, presidents, and other senior business leaders of some of the highest-grossing corporations across industries, a group employing eight million people at the time - in order to assess their beliefs regarding the use and importance of writing.
Once again, the findings were startling: “almost half of the sample offered or required writing training for deficient salaried employees, with an average cost of around $950 per employee.”; “American firms may pay as much as $3.1 billion dollars a year on writing training.” (College Board, 2004, as cited in Aschliman, 2016); and, since the original study excluded government and wholesale trade sectors, the average is likely higher (College Board, 2004, as cited in Aschliman, 2016).
If these are the data of the country with the world’s largest economy and, coincidentally, where English, the global medium of communication, is a native language and the de facto official language, one can easily extrapolate what the state of affairs in other countries, including India, is.
The question at hand is the plausible sequence of causality between the surge in the number of Text Editing Applications, the concomitant and precipitous decline in students’ writing skills, and the deficiency in writing skills of the workforce entering today’s job market. Could it be that our over-reliance on text-editing applications, as mentioned above, is a noteworthy cause of the aforementioned problems? While these software undoubtedly offer a reliable and valuable text editing service to a seasoned writer, do they unwittingly lead a learner to completely and hopelessly depend on them, in a way that impedes his/her growth as a communicator, a writer, and a successful professional in the long run?
According to Psycholinguist Willem Levelt, there are four stages to speech production: conceptualisation, formulation, articulation and self-monitoring. Conceptualisation is the cognitive impulse to say something, and exists at the ‘thought’ level. Once there is an impulse, we enter into the Formulation stage where we begin to frame sentences in our mind, using our knowledge of grammar, vocabulary and composition. We then proceed to Articulation, which is the phonetic expression of the formulated expression. Finally, articulation is complemented with Self-Monitoring, a self-corrective mechanism that refines that expression or utterance.
Sentence formulation involves extensive knowledge of sentence structures and their individual implications; words and their semantics; rules of composition; and much more. Every acceptance of an ‘AI Powered Text Edit’ is an opportunity lost for practicing sentence formulation. Someone who does not rely on such applications would spend a lot of time enhancing his knowledge of the language and then would retrieve, consciously at first but increasingly intuitively as he progresses, the information from memory in order to hone his skill at the language. However, someone who is utterly dependent on these applications would not invest in the acquisition of such knowledge and the development of said skill.
The first such “harmful benefit” of AI Enabled Writing Assistants is the ‘Predictive Text’ feature. Prolonged use of ‘Predictive Text’ has the potential to arrest or shape the development of the cognitive process of Speech Formulation, and leave us utterly and hopelessly incapable of original thought. If allowed to, such applications, quite literally, finish our sentences for us. As we start typing out a sentence, the application starts running its Machine Learning algorithms to analyse and predict what we’re trying to say; and, through a complex combination of situational analysis and linguistic probability, it comes up with a handful of recommendations that, more often than not, are accurate. So much so that you could complete your sentence just by clicking on the suggestions, never having to type a single letter beyond the first word.
Now, although this feature is an excellent ‘time saver’ for those who’re already good at the language, it is counter productive to those who are still learning to write or wish to improve their writing skills and possibly even hurts native users in the long run. Clare Wood, a Nottingham Trent University psychologist, warns that as these AI-powered systems rely on learning from what we have typed in the past. This could introduce errors. “If it detects that particular ungrammatical word combinations frequently co-occur then these will be reinforced,” she adds. According to Ken Arnold, a researcher at Harvard’s school of engineering and applied sciences, "Predictive text systems are starting to offer suggestions that are longer, more coherent, and more contextual than ever before,” Furthermore, a study found that secondary school children who used predictive text on their mobile phones made more spelling errors than non-users.
Autosuggest could not only affect what we want to say but also how we say it. Computer scientists at Harvard University and the Draper Laboratory in Cambridge, Massachusetts, have found that using AI-powered predictive text systems could introduce biases into what people write.
"Machine learning algorithms, which are used to train AI systems on large sets of data, can pick up and amplify biases contained within the data. So, a predictive text system that has been trained using text from positive online reviews might tend to suggest words that are more positive as a result."
The second threat is the Autocorrect functionality, which automatically corrects typographical errors as they are committed. Now, this feature undoubtedly has its benefits, especially if you’re someone who’s bad at typing or are someone with podgy fingers, trying to type in a text message on your touchscreen smartphone. However, when you genuinely and routinely forget to put the apostrophe in contractions such as “can’t, won’t, wouldn’t, shouldn’t, etc.”; or, you haven’t capitalised the first letter of the first word of every sentence or even the letters in an acronym such as ‘MBA, HR, UN, USA, etc’; or, you habitually “misspell” words with double letters in them, such as ‘beginning, Philippines, hobbies, questionnaire, beautifully, misspell, etc.”; or you haven’t yet figured out when the ‘i’ goes before the ‘e’ or vice versa in words such as ‘believe, receive, conceive, fierce, weird, perceive, etc.”; or you’re still confused whether Americans use ‘s’ or ‘z’ in ‘organise, victimise, etc.’, or ‘s’ or ‘c’ in noun form of words such as ‘advice, practice, etc.’; and start relying on AI enabled software to do your writing for you, you’re disabling your natural learning mechanisms and preventing yourself from learning these rules.
Leaving the correction of errors pertaining to spellings, homophones, punctuation, auto-capitalisation, articles, tenses, basic auxiliary verbs etc., which come under the umbrella of ‘Grammatical and Semantic errors’, to be handled by AI enabled softwares has a rather negative impact on a learner’s subconscious mind. It leads the learner to forgetting spellings of words and even to not taking the pains to learn them. According to Kate Heitkamp, an Ohio school teacher, “Frequently, students who use autocorrect rely so much on it to know what word they have spelled that they don’t stop to see if the word is the correct word they wanted.” (McCarthy, 2019). Moreover, “If students don’t have basic spelling skills, the autocorrect doesn’t seem to help since it will give them an incorrect word.”, Heitkamp further states (McCarthy, 2019). And to top it all, the changes autocorrect makes are so quick that, the minute you hit enter or space, the autocorrect word appears with no trace of the original word that was typed. There is no opportunity to compare the wrong and the right versions, and so there’s no chance of learning, whatsoever.
Alternate Word Suggestions
The ‘Alternate Words suggestion’ feature offered by text editing applications makes matters worse. In the famous sitcom ‘FRIENDS’, American actor Matt LeBlanc’s character, Joey Tribbiani, once had to write a letter of recommendation for his friends Monica and Chandler for an adoption agency. In order to sound smart, he used a thesaurus for ‘every word’ in the letter. And so, he ended up writing, “They're humid, prepossessing Homo sapiens with full-sized aortic pumps.” instead of, “They're warm, nice people with big hearts.”, and signs off with, ‘"Baby Kangaroo" Tribbiani’ for Joey Tribbiani.
Learning synonyms without learning their respective connotations, collocations and register can be devastating to vocabulary enhancement. Reliance on suggestions given by text editing applications for alternative words robs us of the need to learn and understand what context a word best fits. These software highlight the fact that we are wrong and provide us alternate suggestions, but leave us confused for they don’t tell us why we are wrong. We understand that the words we used were wrong, but not why they were wrong. And when we know an AI application is readily available to do the job for us, we are directed away from genuine and productive ways of vocabulary enhancement, such as extensive READING! Cynthia Ryan, an associate professor of English at The University of Alabama at Birmingham, states, “Context is key, and neither autocorrect, a thesaurus nor any other kind of resource can be counted on to do the work for the writer.” Bakken, 2014).
The MCQ Syndrome
Another unintended consequence of Students’ growing reliance on text editing software is what we’d like to call the ‘MCQ syndrome’. Private Coaches and Tutors concur that candidates preparing for entrance examinations of the ‘Multiple Choice Question’ or MCQ Format often exhibit a hopeless dependence on answer choices to solve an aptitude question, be it in Quantitative Ability, Logical Reasoning or Verbal Ability. At times, some individuals become so accustomed to reading and eliminating answer choices that they find it impossible to answer a question that didn’t have answer choices. Readymade choices or suggestions disable independent thought. High levels of dependency on choices and suggestions are making people incapable of independent problem solving.
Prolonged use of text editing software can have a similar effect on the learning of a language, and, more importantly, someone trying to improve their writing skills. It can kill creativity, spawning an addiction of sorts that has a crippling effect on a student of the language arts.
All writers have their own unique style of writing or way of expressing themselves. The choice of words, grammatical constructions, neologisms, figures of speech and flavour of writing differ from writer to writer. Take the novelist Roald Dahl for instance. He was well known for inventing words of his own in his stories and novels. It was part of his creative genius and something that made his works stand out. What do you think would have become of his acclaimed works had he relied hopelessly on text editing applications to help him learn how to write?
Loss of Originality
These applications are AI-powered and programmed to work a certain way. Machine Learning works by combining large amounts of data with fast, iterative processing and intelligent algorithms, allowing the software to learn automatically from patterns or features in the data. So, in other words, the kind and the quality of suggestions depend immensely on the type of data processed. Conversely, the suggestions are also limited to the type of data processed. This cookie cutter approach to editing eats away the uniqueness that abounds in every person’s writing. If different writers are made to use the same software, they’d all likely end up with similar outputs, since the software works within the confines of a limited database. Tucci et al. (2020) confirm, “Machine Learning algorithms, which underpin many of the most advanced AI tools, are only as smart as the data they are given in training. Because a human being selects what data is used to train an AI program, the potential for machine learning bias is inherent and must be monitored closely.” Stuart Rubinow, on Quora, extends the argument further by adding, “Such a software is only as good as its algorithm or database of examples. At its best it will give you correct but rather pedestrian writing.” (Rubinow, n.d.)
Freedom of Expression
And what about intentional errors we make? Imagine how annoying it is to have your computer change intended writing into something it’s not. An example of this is when we sometimes type words or phrases of other languages using the English Alphabet, a.k.a the Latin or Roman Alphabet. Another is any neologism. You work hard to invent the perfect portmanteaux, but your system won’t let you type it in, ‘correcting’ it into something else every time you hit the spacebar.
There are also questions on whether these applications can accommodate intentional lexical and structural ambiguity, as was outlined in the previous section. Remember the different puns authors often riddle books with? What would become of them if these applications are allowed to make corrections on those works? Sooner or later, the writer would have to turn the application off. Why should we use applications which we are going to ignore anyway?
Proponents of these software, especially those with vested interests, retort that these suggestions are, after all, just suggestions and that we are not obligated to adopt them. But then they flagrantly support the large corporations that manufacture these software by endorsing the ability of these software to enhance the writing skills and the need for everyone to adopt them.
But why else would we use these applications if not to take any or all of their suggestions at face value? Having outsourced these tasks to AI enabled software, are we going to evaluate the suggestions critically, mulling over whether we should just accept it, or try to find something more suitable? If that were the case, we wouldn't use these applications in the first place. So even though the changes are suggestive, they are pretty much adopted without question, in most of the cases. This is also one of the reasons the features of a built-in dictionary or a library of grammatical rules in these applications remain unused for the most part. Hence, most users' language development remains compromised, and his/her unassisted sentence formations remain not only abecedarian, but also utterly contrived.
To conclude, these text editing applications won’t always be there with us to save the day. Think of instances when we have to voice our thoughts instead of writing them down. What then? Will these applications be able to check the coherence and logic behind our words?
Not just writing skills but proper speech entails good grammar too. If we are making basic idiomatic mistakes in our writing, then it is more than likely to reflect in how we speak too.
Diana Booher states, “If you can’t write your message in a sentence, you cannot say it in an hour.”
Our formulation of ideas using grammar and vocabulary enhances our articulation. And articulation, as we all know, is the key to effective communication.
For students, being able to express their ideas in a clear and logical manner is as important as submitting high quality essays. Although relying heavily on text editing software applications might help them get their message across, it will not enhance their true skill. As a result, despite having produced fairly decent writings, they will not learn how to express themselves, an impediment made visible particularly when not using these applications.
In the words of Jason Fried, founder of Basecamp and author of the book, Rework, "If you are trying to decide among a few people to fill a position, hire the best writer. [His/her] writing skills will pay off. That's because being a good writer is about more than clear writing. Clear writing is a sign of clear thinking. Great writers know how to communicate. They make things easy to understand. They can put themselves in someone else's shoes. They know what to omit. And those are qualities you want in any candidate. Writing is making a comeback all over our society...Writing is today's currency for good ideas." (Moore, 2016).
Text editing software indeed have their fair share of benefits, but there are also many disadvantages that come with them. If a little effort on our part can help us improve our writing and communication skills each step of the way, then should we not go for it, instead of relying heavily on software applications?
It certainly won’t be an easy feat and might take some dedication, but as the American science fiction author Octavia E. Butler opines, “You don't start out writing good stuff. You start out writing crap, and thinking it’s good stuff, and then gradually you get better at it. That’s why I say one of the most valuable traits is persistence.”
Remember, persistence is the key to success, in any venture, and continued hard work towards improving your grammar and writing skills, too, is sure to pay off in the long run.
Aschliman, C. (2016). Write to Work: The Use and Importance of Writing as Perceived by Business Leaders [Doctoral dissertation, Virginia Commonwealth University]. VCU Scholars Compass. https://doi.org/10.25772/QMNR-5231
Bakken, J. (2014, May 19). Could texting and autocorrect affect kids’ writing skills?. UAB News. Retrieved from https://www.uab.edu/news/youcanuse/item/4697-could-texting-and-autocorrect-affect-kids-writing-skills
Bronowicki, K. A. (2014). Technology’s Adverse Effects on Student’s Writing: An Emphasis on Formal Writing is needed in an Academic Curriculum [Master’s Thesis, State University of New York]. Digital Commons @Brockport. https://digitalcommons.brockport.edu/ehd_theses/392
Chapter 1: Language Production. Tesis Doctorals en Xarsa. (n.d.). https://www.tdx.cat/bitstream/handle/10803/1662/01.CHAPTER_1.pdf?sequence=2&isAllowed=y
Deighton, L. (2013, August 16). Teaching spelling and grammar is still vital in a technology-driven classroom. The Guardian. Retrieved from https://www.theguardian.com/teacher-network/teacher-blog/2013/aug/16/teaching-literacy-technology-ict-schools
McCarthy, A. (2019, August 13). How 'smart' email could change the way we talk. BBC. Retrieved from https://www.bbc.com/future/article/20190812-how-ai-powered-predictive-text-affects-your-brain
Moore, K. (2016, April 7). Study: 73% of Employers Want Candidates With This Skill. Inc.. Retrieved from https://www.inc.com/kaleigh-moore/study-73-of-employers-want-candidates-with-this-skill.html
Parker, M. (2018, October 16). Grammarly helps and hinders students’ writing. The Feather Online. Retrieved from https://www.thefeather.com/2018/10/16/grammarly-helps-and-hinders-students-writing/
Patil, S. (2017, August 10). Language Models: Spellchecking and Autocorrection. Towards data science. Retrieved from https://towardsdatascience.com/language-models-spellchecking-and-autocorrection-dd10f739443c
Purcell, K., Buchanan, J., & Friedrich, L. (2013, July 16). The Impact of Digital Tools on Student Writing and How Writing is Taught in Schools. Pew Research Center. Retrieved from https://www.pewresearch.org/internet/2013/07/16/the-impact-of-digital-tools-on-student-writing-and-how-writing-is-taught-in-schools/
Rubinow, S. (n.d.). Can we trust the grammarly apps?. Quora. Retrieved from https://www.quora.com/Can-we-trust-the-grammarly-apps/answer/Stuart-Rubinow?ch=3&share=1f64900e&srid=uQqTw
Tucci, L., Burns, E., & Laskowski N. (2020, November). artificial intelligence. Search Enterprise AI. Retrieved from https://searchenterpriseai.techtarget.com/definition/AI-Artificial-Intelligence